Original Article
Toll-like receptors promote apoptosis of retinal ganglion
cells in diabetic retinopathy via regulating Caspase-3
and Bcl-xL
Lili Wang1*, Qingxiang Ma2*
1Department of Ophthalmology, Yulin Xing Yuan Hospital, Yulin, Shaanxi, China; 2Department of Ophthalmology,
The First Hospital of Yulin, Yulin, Shaanxi, China. *Equal contributors.
Received September 5, 2015; Accepted December 1, 2015; Epub January 1, 2016; Published January 15, 2016
Abstract: As a complication of diabetes, diabetic retinopathy (DR) is a leading cause of poor vision and blindness. In order to figure out the pathogenesis of DR, this study started with the expression of Toll-like receptor 2 (TLR2) and TLR4 in retinas of diabetic mice induced by streptozotocin in vitro. RGCs were isolated from wild-type (WT) mice and TLR-knockout (TLR-/-) mice, and cultured in high glucose conditions or normal conditions. Cell viability assay and
apoptosis assay were performed to analyze the influences of TLRs on RGC viability and apoptosis. The regulatory mechanisms of TLRs and apoptosis-related factors, cysteine-aspartic acid protease-3 (Caspase-3) and the long iso -form of B-cell CLL/lymphoma 2-like 1 (Bcl-xL), were further analyzed. Results showed up-regulated expression levels of TLR2 and TLR4 in retinas of diabetic mice (P < 0.05). RGCs from WT mice under high glucose conditions were suppressed in viability and promoted in apoptosis compared to RGCs cultured in normal conditions (P < 0.05). While the viability or apoptosis of RGCs from TLR-/- mice was not affected by high glucose culturing conditions. Related
regulatory analyses revealed TLRs could up-regulate Caspase-3 and down-regulate Bcl-xL (P < 0.05). Furthermore, the promoted expression of Caspase-3 and the suppressed expression of Bcl-xL were detected in RGCs from WT mice under high glucose conditions, but not in RGCs of normal conditions, or RGCs from TLR-/- mice. These data
implied that TLRs promoted apoptosis of RGCs via regulating Caspase-3 and Bcl-xL, which might be a pathogenic mechanism of DR.
Keywords: Diabetic retinopathy, retinal ganglion cell, Toll-like receptor, apoptosis
Introduction
Diabetic retinopathy (DR) is a retina disease caused by diabetes. It affects a considerable part of diabetic patients and can eventually lead to blindness [1, 2], thus becoming the leading cause of blindness for middle-aged
people in the United States [3]. The progres -sion of DR can be divided into two stages, namely non-proliferative DR and proliferative DR. No typical sign or symptom is visible to the patients of non-proliferative DR, though macu-lar edema may begin to occur during this stage [4]. But when DR progresses to the next stage,
bleeding occurs in the eye and makes the eye -sight of patients even worse [5]. Numerous studies have discussed the pathogenesis of DR. Potential regulation targets of DR have been studied, such as intercellular adhesion
molecule 1 and its leukocyte counter-receptor
CD18 [6], placental growth factor-1 [7], and
vascular endothelial growth factor [8]. The num -ber reduction and pathological changes of reti-nal ganglion cells (RGCs) have been found in DR [9], and researchers detect nuclear transloca-tion of GAPDH in these RGCs [10]. However, the pathogenesis of DR, especially the regulatory mechanism of RGCs, is still unclear.
Family of Toll-like receptor (TLR) consists of
members vital for activating innate immune
responses [11]. The pivotal roles of TLRs in reg
-ulating viral pathogenesis make them promis -ing therapeutic targets for treat-ing immune
were experimentally induced by streptozotocin for detecting TLR2 and TLR4 expression. RGCs were isolated from wild-type (WT) mice and
TLR-knockout (TLR-/-) mice and cultured in high
glucose conditions to analyze changes in cell
viability and apoptosis in vitro. Further, the
reg-ulatory relationships between TLRs and apop -tosis-related factors, cysteine-aspartic acid protease-3 (Caspase-3) and the long isoform
of B-cell CLL/lymphoma 2-like 1 (Bcl-x), Bcl-xL, were discussed. These findings will provide
valuable information on the regulatory mecha-nisms of RGCs and the pathogenesis of DR.
Material and methods
Animals
Clean grade WT C57BL/6J mice (The Jackson
Library, Bay Harbor, ME, USA) weighing from 20 to 25 g were randomly grouped into the nondia-betic control group and the experimentally induced diabetic group (six individuals for each
group). To generate diabetic mice, freshly pre
-pared streptozotocin (Sigma-Aldrich, Shanghai,
China) dissolved in citrate buffer (pH 4.5) was
intraperitoneally injected (60 mg/kg) to mice that had been deprived of food for 12 h. The
injection was performed once every day for 5 d.
To keep the mice insulin deficient but not cata -bolic, neutral protamine hagedorn (NPH) insulin
(Wanbang, Xuzhou, China) of 0.1 units was subcutaneously injected once per week. Then
the mice were raised with free access to food and water. Food consumption, body weight and glycated hemoglobin (Bio-swamp, Wuhan,
China) were measured every week to estimate
the level of hyperglycemia. Retinas of these mice were sampled for expression analyses.
TLR-/- mice were purchased from Biomodel
Organism (Shanghai, China). The experiments
were approved by a local animal committee for
supplemented with 10% FBS, 1 mg/mL
glu-cose, 100 U/mL penicillin, 100 μg/mL strepto
-mycin and 2 μg/mL amphotericin
(Sigma-Aldrich). Cells were counted and adjusted to the concentration of 1×105/mL, and cultured on plates coated with mouse tail collagen
(Sybio, Hangzhou, China). After 24 h, 5-Bromo-2’-deoxyuridine (20 μg/mL, Sigma-Aldrich) was
added to inhibit the growth of non-neuronal cells, and the medium was changed after 48 h. Cells were cultured at 37°C with 5% CO2. RGCs
were identified by morphology and the mono
-clonal anti-THY1 antibody (Sigma-Aldrich). For RGCs in normal conditions, RGCs from six WT
mice or six TLR-/- mice were designated as
RGC (WT) and RGC (TLR-/-), respectively. For RGCs in high glucose conditions, RGCs form
six WT mice or six TLR-/- mice were cultured in DMEM (high glucose) with 4.5 mg/mL
glu-cose, designated to be RGC (WT) +HG and RGC
(TLR-/-) +HG, respectively.
Real-time quantitative PCR (qRT-PCR)
Total RNA was extracted from retinas using Trizol (Invitrogen, Carlsbad, CA, USA), and
re-verse transcription was performed by Primer-
Script RT Reagent Kit (TaKaRa, Dalian, China) according to the manuals. Each 20 μL PCR
reaction contained 20 ng complementary DNA
and the specific primers for mouse TLR2
(for-ward: 5’-GAT AAT GAA CAC CAA GAC CTA CC-3’ and reverse: 5’-GCA GTT CTC AGA TTT ACC
CA-3’), TLR4 (forward: 5’-ACT CTG ATC ATG GCA CTG TTC TT-3’ and reverse 5’-GCT CAG ATC TAT GTT GGT TGA-3’) or GAPDH (forward: 5’-TCA ACA GCA ACT CCC ACT CTT CCA-3’ and reverse: 5’-ACC CTG TTG CTG TAG CCG TAT TCA-3’).
Re-actions were conducted on LightCycler 96
Real-Time PCR System (Roche, Basel, Switzerland)
com-prised of denaturation at 94°C for 1 min, annealing at 60°C for 1 min and extension at
72°C for 1 min, and a step of final extension at
72°C for 7 min. All samples were tested in trip-licate. GAPDH was used as the internal control and data were calculated by the 2-ΔΔCt method.
Western blot
Protein samples of 20 μg from retinas and
RGCs were denatured at 100°C for 5 min, sep-arated by sodium dodecyl sulfate-polyacryl-amide gel electrophoresis (SDS-PAGE), and
blotted to polyvinylidene fluoride membrane. The membrane was blocked in 5% skim milk
for 2 h at room temperature and incubated in
the primary antibody overnight at 4°C. The pri -mary antibodies were mouse monoclonal
anti-TLR2, TLR4, Caspase-3 or Bcl-xL (Invitrogen), and GAPDH was the internal reference. Then
the blot was incubated in the horseradish per-oxidase-conjugated secondary antibody for 1 h at room temperature. Signals were detected by enhanced chemiluminescence reagent and
analyzed by ChemiScope 3600 Mini (CLINX,
[image:3.629.103.529.79.477.2]Shanghai, China).
Cell viability assay
RGCs were seeded on the 96-well plate and adjusted to the concentration of 1×104/well.
The cells were cultured at 37°C for 24 h. Then cell viability was analyzed using 3-(4,5-dimeth
-ylthiazol-2-yl)-2,5-dephenyltetrazolium bromide (MTT) Cell Proliferation and Cytotoxicity Assay Kit (Beyotime, Shanghai, China) according to the manuals. MTT (5 mg/mL) was added to the medium. After 4 h of incubation, formanzan dis
-solving solution (100 μL) was add to each well and the plate was shaken slowly for 4 h. Then
the absorbance was detected at 570 nm with a
spectrophotometer (Thermo Fisher Scientific,
Waltham, MA, USA).
Cell apoptosis assay
Terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick-end labeling (TUNEL) was
conducted using the In Situ Cell Death De-
tection Kit (Roche Diagnostics, Mannheim, Germany) to analyze cell apoptosis. Cell smears were prepared and fixed using Immunol Stain-ing Fix Solution (Beyotime). TUNEL reaction mixture (50 μL) was added, and the slides were incubated in a humidified atmosphere for 60 min at 37°C in the dark. Then the slides were observed with a microscope and five view fields were randomly selected for each sample to count the apoptotic (TUNEL-positive) cell
number.
Statistical analysis
Statistical analysis was performed by Statistical Product and Service Solutions (SPSS) 19 (IBM,
New York, USA). The significance of difference between groups of qRT-PCR, western blot, MTT and TUNEL experiments was tested by t test, before which F test was performed to test the homogeneity of variance. Differences were
considered significant if P < 0.05.
Results
TLR2 and TLR4 are up-regulated in RGCs of DR
We first compared the expression of TLR mRNA and proteins in RGCs between the WT control and the diabetic mouse induced by streptozoto -cin. Both TLR2 and TLR4 mRNA levels detected
by qRT-PCR were significantly up-regulated in
retinas of diabetic group compared to the con-trol group (P < 0.05, Figure 1A and 1B). Besides,
the protein expression levels of TLR2 and TLR4
detected by western blot were also increased
significantly in the diabetic group (P < 0.05,
Figure 1C and 1D). These results indicated that
TLR2 and TLR4 were up-regulated in the dia -betic retinas, implying their potential roles in regulating DR.
TLRs inhibit viability and promote apoptosis of RGCs in DR
After the RGCs were cultured in high glucose conditions or normal conditions, we conducted
cell viability assay with MTT method to analyze the relationship between TLRs and RGCs. Re-sults showed the cell viability of RGC (WT) +HG was obviously inhibited compared to RGC (WT)
(P < 0.05, Figure 2A), indicating high glucose conditions could lead to inhibition of RGC viabil-Figure 2. Viability changes of RGCs in high glucose conditions compared to normal conditions. A. Viability changes of RGCs from WT mice. B. Viability changes of RGCs from TLR-/- mice. Cell viability was detected at 0 h, 72 h and 7 d
[image:4.629.100.529.81.224.2]ity. But the viability of RGCs from TLR-/- mice was not affected by high glucose, with no
sig-nificant difference between RGC (TLR-/-) and RGC (TLR-/-) +HG (P > 0.05, Figure 2B). These
data revealed that TLRs were closely related to
the reduced RGC viability in DR.
Similarly, apoptosis of RGCs was analyzed by TUNEL after cells were treated by high glucose for 3 d. The apoptosis of RGC (WT) +HG was significantly promoted by the high glucose con
-ditions compared to RGC (WT) (P < 0.05, Figure 3A). While the apoptosis showed no significant
difference between RGC (TLR-/-) +HG and RGC (TLR-/-) (P > 0.05, Figure 3B). So TLRs might play roles in inducing the RGC apoptosis in DR.
TLRs mediate RGCs via regulating Caspase-3 and Bcl-xL
Since TLRs might mediate cell apoptosis in DR, we further analyzed its regulation on two apop -tosis-related factors, Caspase-3 and Bcl-xL.
Caspase-3 was significantly up-regulated in RGC (WT) +HG compared to RGC (WT) (P < 0.05, Figure 4A and 4B). But it was significantly inhibited when TLRs were knocked out (P < 0.05). While RGCs were cultured in normal
con-ditions, it seemed that knockout of TLRs did
not affect the expression of Caspase-3. Similar experiments were conducted on Bcl-xL (Figure 4A and 4C). RGC (WT) +HG expressed Bcl-xL in a significant lower level compared to RGC (WT)
(P < 0.05). In normal culture conditions, the expression levels of Bcl-xL were not affected
when TLRs were knocked out (P > 0.05). But in
high glucose conditions, Bcl-xL was
up-regulat-ed by the knockout of TLRs (P < 0.05). Taken
together, the expression levels of Caspase-3 and Bcl-xL were up-regulated and down-regu-lated, respectively, when RGCs were cultured in high glucose conditions. In normal culturing conditions, the expression of Caspase-3 or
Bcl-xL was not affected by TLRs. However, in high glucose conditions, knockout of TLRs could
inhibit Caspase-3 and promote Bcl-xL. It could be deduced from these results that Caspase-3 and Bcl-xL participated in the regulation of
RGCs in DR, and that TLRs could mediate RGCs
via up-regulating Caspase-3 and
down-regulat-ing Bcl-xL. The regulatory functions of TLRs,
Caspase-3 and Bcl-xL on RGC apoptosis might be pivotal to the pathogenesis of DR.
Discussion
It has been postulated that TLR2 and TLR4 are
associated with eye diseases induced by diabe-tes [14, 17]. In the present study, we use
strep-tozotocin-induced diabetic mice and detect the high expression levels of TLR2 and TLR4 in reti -nas. In vitro experiments on RGCs from WT and
TLR-/- mice indicate TLRs inhibit the viability and
promote the apoptosis of RGCs in DR. Then
experiments on cell apoptosis-related factors show TLR-knockout down-regulates Caspase-3
and up-regulates Bcl-xL in RGCs under high
glucose condition, implying TLRs mediate the
apoptosis of RGCs in DR via regulating Cas- pase-3 and Bcl-xL. But in normal RGCs, Cas-
[image:5.629.103.530.82.226.2]pase-3 and Bcl-xL are not affected by TLRs, suggesting the regulation of RGCs by TLRs,
Figure 3. Apoptosis changes of RGCs in high glucose conditions compared to normal conditions. A. Apoptosis chang-es of RGCs from WT mice. B. Apoptosis changchang-es of RGCs from TLR-/- mice. Detection was conducted at 3 d post the
Caspase-3 and Bcl-xL may be significant to DR pathogenesis. These data reflect the associa
-tion between TLRs and RGC apoptosis in DR,
providing possible therapeutic targets for DR treatment.
Caspase-3 and Bcl-xL, the two cell apoptosis-related factors used in this study, are capable of regulating cell apoptosis. Caspase-3 could
be activated by Caspase-9, participating in effi -cient execution of cell apoptosis [18]. Active Caspase-3 has been used for detecting apop-tosis induced by a wide variety of apoptotic sig-nals [19]. Its roles in regulating RGCs have been reported [20-22]. Inhibition of Bcl-2 and Bcl-xL increases the apoptosis of human myeloid
leu-kemia cells [23]. The possible functions of
xL in RGCs have also been studied. Loss of
Bcl-xL significantly increase the death rate of RGCs
after axonal injury, implying its anti-apoptosis roles in RGCs [24]. Results of this study
sug-gested that TLRs inhibited viability and promot -ed apoptosis of RGCs, which was achiev-ed by up-regulating Caspase-3 and down-regulating
Bcl-xL. These results were consistent with pre -vious studies, indicating the pro-apoptosis roles of Caspase-3 and the anti-apoptosis roles of Bcl-xL. Additionally, the altered expression levels of Caspase-3 and Bcl-xL were only de- tected in high glucose conditions, so apoptosis
of RGCs regulated by TLRs, Caspase-3 and
Bcl-xL seemed pivotal to DR pathogenesis.
TLRs can regulate cell apoptosis via a variety of apoptosis-related factors. TLR2 is capable of
accelerating the apoptosis of human monocytic
cell line THP-1 via inhibiting p38 mitogen-acti
-vated protein kinase and tumor necrosis factor-α [25]. In myelomonocytic leukaemia cells, TLR2 inhibits the transcriptional factor NF-кB
and suppresses apoptosis of tumor cells [26].
[image:6.629.103.526.79.388.2]With regard to TLR4, it mediates apoptosis of
Figure 4. TLRs regulating Caspase-3 and Bcl-xL in RGCs under high glucose conditions. A. Western blot showing Cas -pase-3 and Bcl-xL were regulated by TLRs in RGCs under high glucose conditions. B. Histogram showing the relative signal intensity of Caspase-3 in the Western blot results. C. Histogram showing the relative signal intensity of Bcl-xL in the western blot results. *, Significant differences compared to the other groups (P < 0.05). RGC, retina ganglion cell. WT, wild-type. HG, high glucose. TLR, Toll-like receptor. TLR-/-, TLR-knockout. Caspase-3, cysteine-aspartic acid
intestinal epithelial cells via cyclooxygenase 2 [27], and suppresses Wnt signaling to regulate apoptosis of photoreceptors [28]. In DR, the
TLR-mediated NF-кB activation is further regu
-lated by microRNA-146 [29]. Therefore, TLRs
and multiple regulatory factors composing the complex regulatory parts of cell apoptosis.
Though there are studies indicating down-regu
-lation of TLR4 and Caspase-3 suppresses neu -ronal apoptosis [30], no explicit evidence re-
fers to the associations between TLRs and
Caspase-3 or Bcl-xL. In this study, we
discov-ered the regulatory functions of TLRs on
Cas-pase-3 and Bcl-xL, providing new references for RGC apoptosis and DR pathogenesis.
To sum up, this study detects higher expression levels of TLR2 and TLR4 in retinas of diabetic
mice and uncovers the anti-viability and
pro-apoptosis roles of TLRs. The regulatory func
-tions of TLRs in RGC apoptosis of DR patients
are executed via mediating Caspase-3 and
Bcl-xL. These findings contribute to the elucidation
of DR pathogenesis and provide possible thera-peutic targets for DR treatment.
Acknowledgements
This research was conducted in Department of Ophthalmology, Yulin Xing Yuan Hospital, and
there was no foundation to support.
Disclosure of conflict of interest
None.
Address correspondence to: Dr. Lili Wang, Depart- ment of Ophthalmology, Yulin Xing Yuan Hospital, 33 West Renmin Road, Yulin 719000, Shaanxi, China. Tel: +86-0912-3237408; E-mail: wanglili3331@126. com
References
[1] Jones CD, Greenwood RH, Misra A and Bach- mann MO. Incidence and progression of dia-betic retinopathy during 17 years of a popula-tion-based screening program in England. Diabetes Care 2012; 35: 592-596.
[2] Xu J, Xu L, Wang YX, You QS, Jonas JB and Wei WB. Ten-year cumulative incidence of diabetic retinopathy. The Beijing Eye Study 2001/2011. PLoS One 2014; 9: e111320.
[3] Engelgau MM, Geiss LS, Saaddine JB, Boyle JP, Benjamin SM, Gregg EW, Tierney EF, Rios-Burrows N, Mokdad AH, Ford ES, Imperatore G
and Narayan KM. The evolving diabetes bur -den in the United States. Ann Intern Med 2004; 140: 945-950.
[4] Ramalho LH, Avila MP, Moraes Júnior HV and Damasceno Ede F. Subclinical diabetic macu-lar edema and mild non-proliferative diabetic retinopathy: data correlation with the retinal thickness analyzer (RTA). Arq Bras Oftalmol 2009; 72: 503-508.
[5] Celik T, Berdan ME, Iyisoy A, Kursaklioglu H, Turhan H, Kilic S, Gulec M, Ozturk S and Isik E. Impaired coronary collateral vessel develop-ment in patients with proliferative diabetic reti-nopathy. Clin Cardiol 2005; 28: 384-388. [6] Joussen AM, Poulaki V, Le ML, Koizumi K,
Esser C, Janicki H, Schraermeyer U, Kociok N, Fauser S, Kirchhof B, Kern TS and Adamis AP. A central role for inflammation in the patho -genesis of diabetic retinopathy. FASEB J 2004; 18: 1450-1452.
[7] Miyamoto N, de Kozak Y, Jeanny JC, Glotin A, Mascarelli F, Massin P, BenEzra D and Behar-Cohen F. Placental growth factor-1 and epithe-lial haemato-retinal barrier breakdown: poten -tial implication in the pathogenesis of diabetic retinopathy. Diabetologia 2007; 50: 461-470. [8] Myśliwiec M, Balcerska A, Zorena K, Myśliwska
J, Lipowski P and Raczyńska K. The role of vas -cular endothelial growth factor, tumor necrosis factor alpha and interleukin-6 in pathogenesis of diabetic retinopathy. Diabetes Res Clin Pract 2008; 79: 141-146.
[9] Meyer-Rüsenberg B, Pavlidis M, Stupp T and Thanos S. Pathological changes in human retinal ganglion cells associated with diabetic and hypertensive retinopathy. Graefes Arch Clin Exp Ophthalmol 2007; 245: 1009-1018. [10] Shimada A, Fujikawa A, Miyata Y and Kosano
H. Nuclear translocation of GAPDH in retinal ganglion cells in an early event in diabetic reti-nopathy. J Endocrinol Diabetes Obes 2014; 2: 1058.
[11] Kumar H, Kawai T and Akira S. Toll-like recep -tors and innate immunity. Biochem Biophys Res Commun 2009; 388: 621-625.
[12] Lester SN and Li K. Toll-like receptors in antivi -ral innate immunity. J Mol Biol 2014; 426: 1246-1264.
[13] Oh JY, Choi H, Lee RH, Roddy GW, Ylostalo JH, Wawrousek E and Prockop DJ. Identification of the HSPB4/TLR2/NF-kappaB axis in macro -phage as a therapeutic target for sterile inflam -mation of the cornea. EMBO Mol Med 2012; 4: 435-448.
Ophthalmol 2015; 253: 389-398.
[18] Brentnall M, Rodriguez-Menocal L, De Guevara RL, Cepero E and Boise LH. Caspase-9, cas-pase-3 and caspase-7 have distinct roles dur-ing intrinsic apoptosis. BMC Cell Biol 2013; 14: 32-40.
[19] Mazumder S, Plesca D and Almasan A. Cas-pase-3 activation is a critical determinant of genotoxic stress-induced apoptosis. Methods Mol Biol 2008; 414: 13-21.
[20] Tchedre KT and Yorio T. sigma-1 receptors protect RGC-5 cells from apoptosis by regulat-ing intracellular calcium, Bax levels, and cas-pase-3 activation. Invest Ophthalmol Vis Sci 2008; 49: 2577-2588.
[21] Wang S and Li K. MicroRNA-96 regulates RGC-5 cell growth through caspase-dependent apoptosis. Int J Clin Exp Med 2014; 7: 3694-3702.
[22] Li GY, Fan B and Su GF. Acute energy reduction induces caspase-dependent apoptosis and ac-tivates p53 in retinal ganglion cells (RGC-5). Exp Eye Res 2009; 89: 581-589.
[23] Rahmani M, Aust MM, Attkisson E, Williams DC Jr, Ferreira-Gonzalez A and Grant S. Dual inhibition of Bcl-2 and Bcl-xL strikingly enhanc -es PI3K inhibition-induced apoptosis in hu-man myeloid leukemia cells through a GSK3- and Bim-dependent mechanism. Cancer Res 2013; 73: 1340-1351.
Promotes Tumor Progression in Vivo. Acta Naturae 2011; 3: 83-93.
[27] Fukata M, Chen A, Klepper A, Krishnareddy S, Vamadevan AS, Thomas LS, Xu R, Inoue H, Arditi M, Dannenberg AJ and Abreu MT. Cox-2 is regulated by Toll-like receptor-4 (TLR4) sig -naling: Role in proliferation and apoptosis in the intestine. Gastroenterology 2006; 131: 862-877.
[28] Yi H, Patel AK, Sodhi CP, Hackam DJ and Hackam AS. Novel role for the innate immune receptor Toll-like receptor 4 (TLR4) in the regulation of the Wnt signaling pathway and photoreceptor apoptosis. PLoS One 2012; 7: e36560.
[29] Cowan C, Muraleedharan CK, O’Donnell JJ, Singh PK, Lum H, Kumar A and Xu S. MicroRNA-146 inhibits thrombin-induced NF-kappaB activation and subsequent inflamma -tory responses in human retinal endothelial cells. Invest Ophthalmol Vis Sci 2014; 55: 4944-4951.
